The production of silicon wafers for integrated circuit substrates requires the use of complex processes involving many mechanical and chemical steps. Achieving a high degree of purity of the ultimate product is one of the most important concerns in the process, because the presence of too high a concentration of contaminants in or on a wafer may adversely affect the performance of circuits fabricated on that wafer. Beginning with a raw material such as quartzite, the impurity level of the silicon is typically reduced by eight or more orders of magnitude before the final silicon wafer is ready for shipment. Quartzite has relatively high impurity levels; for instance, it may contain aluminum levels of approximately 3.times.10.sup.20 atoms/cm.sup.3. Through a series of chemical steps, this concentration may be reduced to the 10.sup.12 range. The final ultra-pure product is referred to as electrical grade polycrystalline silicon (EGS), and may be formed into a single crystal ingot from which wafers are cut. Once this high purity has been achieved, it is important that the purity is not compromised in the later wafer cutting, shaping and polishing processes.
The mechanical and chemical steps involved in the wafer production process present many opportunities for the introduction of impurities onto the surfaces of the wafers. A typical process is as follows: Before wafers are cut from the ingot, the ingot itself is shaped. Industrial grade diamond grinding wheels are used to form the ingot into a uniform cylindrical shape. A similar grinding tool is then used to introduce a primary flat along one side of the ingot. This flat is used by later processing equipment to align wafers in the equipment and is also used as a crystallographic reference. Next individual wafers are cut from the ingot using a stainless steel blade with diamond particles bonded to the edge. Some degree of inconsistency exists in the thickness of the newly cut wafers, so a lapping and grinding step is used to correct variations in the thickness of the wafers. This step uses a polishing compound, such as an Al.sub.2 O.sub.3 and glycerin mixture, with which the wafer is mechanically reduced to the correct thickness. After these machining steps, the surface may contain some degree of damage and contamination, so a wet etching step is used to clean and smooth the surface. Typically, a solution of nitric, hydrofluoric and acetic acids is used as the etching solution. After etching, a polished surface is created on one side of the wafer through a mechanical-chemical process that uses a slurry of very fine silica powder and sodium hydroxide. The process may end here, or epitaxial layers may be applied to the surface of the silicon wafer. To ensure wafer quality throughout the manufacturing process, the purity of the wafer surfaces may be measured after each processing step. The final wafer ideally should have a surface contamination concentration of no more than approximately 10.sup.10 impurity atoms/cm.sup.2.
Several common techniques exist for measuring the purity of wafer surfaces. One widely used technique is known as total reflection x-ray fluorescence spectroscopy, or TXRF. TXRF is a nondestructive technique that measures the x-ray fluorescence from surface impurity atoms resulting from excitation caused by an incident x-ray beam. Because it is non-destructive, it can be used to test wafers that are in production. Another common technique is Vapor Phase Decomposition Graphite Furnace Atomic Absorption spectroscopy (VPDGFAAS). In this technique, a chemical treatment is used to decompose the surface layer of a wafer, after which a sample is extracted from the decomposed layer and vaporized in a graphite furnace for standard atomic absorption analysis.
These tests use machines that must be calibrated before quantitative measurements may be made. The calibration process serves various purposes. For TXRF, the calibration process is used to verify that the instrument has not gone out of calibration since the last measurements were made. When used to quantitatively analyze an experimental wafer for contaminants, TXRF techniques typically make a measurement directly on the wafer surface. Any standard used to calibrate this measurement should be of a similar sample type, and should fit in the same machine as the experimental samples. Therefore, a good standard to use for TXRF machine calibration when performing measurements on silicon wafers is an intentionally contaminated silicon wafer.
Intentionally contaminated silicon wafers have other experimental uses as well. For instance, the contaminated wafers may be used with techniques such as VPDGFAAS to verify that the sample collection process is reproducibly extracting samples for measurements. This may be done by first performing the sample extraction technique on an intentionally contaminated wafer, measuring the extracted impurity levels, and then comparing the experimental results to the known contamination level of the wafer. Another example of a use for intentionally contaminated wafers is in gettering experiments. During the processing of integrated circuits, unintentional contamination can occur in many different steps of production. Gettering techniques are used to draw these contaminants away from working areas of the circuits. To obtain accurate quantitative information on the effectiveness of a gettering technique, experiments must be performed on intentionally contaminated wafers with known levels of impurities to measure the quantity of impurities drawn away from the surface by the technique.
Yet another example of a use for intentionally contaminated wafers is in the study of the effects of contaminants on circuit performance. To maximize efficiency in these applications, it is desirable to have wafers with controlled concentrations of more than one impurity. This also allows the study of the effects of multi-elemental contamination on various circuit performance characteristics or fabrication processes.
A common problem that arises in the production of intentionally contaminated wafers concerns the distribution of contaminants on the surface of a wafer. It is desirable to have a high degree of uniformity of contaminant concentrations on the surface to reduce the dependency of a given measurement technique on the position of a detector. Some techniques for contaminating wafers involve dripping a very small quantity of solution containing a known or calculable number of impurity atoms onto the surface of a wafer and then drying the solution. These techniques result in small, localized areas of contamination, and require much care to be taken in positioning a detector when measuring the contamination. The ambient chosen for sample drying can further change the distribution of atoms on the surface due to thermal migration or diffusion, thereby decreasing the predictability and reproducibility of the techniques. These techniques are also not suitable for making wafers for gettering or circuit design studies, as the contamination exists only on small, discrete areas of the surface.
Other techniques involve exposing substantially all of a wafer surface to a contaminant solution and drying the contaminants onto the surface. These techniques reduce the position sensitivity problems, but other problems arise with the controllability of the concentration of contaminants that adsorb to the surface. The amount of contaminant that will adsorb to a wafer surface is dependent upon the interaction between the wafer surface and the contaminant atoms, and may not always be varied by simply changing the concentration of contaminants in the solution.
Many contaminants adsorb to the wafer surface by physiadsorption, a low energy interaction between a contaminant atom on the surface of a wafer and the surface atoms of the wafer itself. Physiadsorption is readily reversible, and occurs when local areas of electrical polarization allow a weak electrostatic attraction to occur between the surface and the contaminant. The quantity of atoms that physiadsorb to a surface may be a function of the number of contaminant species in the solution, the concentration of each species, the free energy of each adsorption reaction, and the effect the solution has on the surface properties of the wafer. Thus, the concentration of contaminant species that adsorbs to a wafer surface is not necessarily a function only of the concentration of contaminants in the solution. Increasing or decreasing the contaminant level in the solution may have little or no effect on the resulting surface contamination level. As a result, problems are encountered both in controllably contaminating wafer surfaces, and in identifying the correct procedure to follow to determine the best solution composition.